"Would you prefer that one person be horribly tortured for fifty years without hope or rest, or that 3^^^3 people get dust specks in their eyes?
I think the answer is obvious. How about you?"
Yes, Eliezer, the answer is obvious. The answer is that this is a false dilemma, and that I should go searching for the third alternative, with neither 3^^^3 dust specks nor 50 years of torture. These are not optimal alternatives.
Actually, there can be multi-level selection (MLS theory; cf. http://en.wikipedia.org/wiki/Group_selection#Multilevel_selection_theory) when there is competition between groups. In the same sense there is selection between individuals when there is competition between individuals, or the competition between genes popularized by Richard Dawkins.
http://www.americanscientist.org/my_amsci/restricted.aspx?act=pdf&id=16386020847008http://www.americanscientist.org/my_amsci/restricted.aspx?act=pdf&id=16386020847008 is a good primer.
This is the best solution for Darwin's problem of ant colonies, even better than haplodiploidy. I thought I would come out of lurking while reading through the sequences to mention this, since multi-level selection was demonized during the 70s under the name "group selection" due to some overzealous proponents. So, while we would not say "evolution has favored a species that buys lottery tickets", we might hypothesize evolution favors human societies that buy lottery tickets when under competition with other societies that do not (as an example).